The coursework part of the compulsory modules provides knowledge of the core methods of biostatistics and gives a first experience in applying and extending these. Lectures are in general completed by exercises (LE) and hours per week are divided into lecture and exercise part (e.g. 2+1). The following table shows the current requirements.

Likelihood Inference | LE 2+1 | 5 CP | |

Generalized Regression | LE 2+1 | 5 CP | |

Clinical Biostatistics | LE 2+1 + 1 lab | 6 CP | |

Good Statistical Practice | flipped classroom | 4 CP | |

Statistical Methods in Epidemiology | LE 2+1 | 5 CP | |

Survival Analysis | LE 2+1 (half semester) | 3 CP | |

Biostatistics Journal Club | Seminar | 4 CP | |

Statistical Consulting | Project | 6 CP | |

Master's Thesis | 30 CP | ||

Master Exam | 3 CP | ||

Overview over the basics of statistical inference. Topics include the introduction to the concept of likelihood and the discussion of likelihood functions of a large variety of statistical models, sufficiency and the likelihood principle, properties of maximum likelihood estimates, standard errors, confidence intervals and pivots, score function and Fisher information, Cramer-Rao bound, confidence intervals and significance tests based on the Wald, score and likelihood ratio statistic, variance-stabilizing transformations, treatment of nuisance parameters, conditional and profile likelihood. The lecture also covers the very basic ideas of Bayesian statistics.

- Held, L. and Sabanés Bové, D. (2013): Applied Statistical Inference: Likelihood and Bayes, Springer.

Introduction to modern regression methods. After a brief recap of classical regression techniques the following topics will be discussed: exponential family of distributions and generalized linear models (GLM), estimation and inference for GLMs, likelihood ratio and deviance, normal linear models, Categorical data and logistic regression, Poisson regression and log-linear models. If time allows we might also discuss mixed effects models, nonparametric regression and additive models.Textbooks are

- Fahrmeir, L., Kneib, T. and Lang, S. (2013), Regression: Models, Methods and Applications, Springer.

- Tutz, G. (2012). Regression for Categorical Data, Cambridge University Pres.

Uncertainty is the rule in medicine and the science of managing medical uncertainty is biostatistics. The aim of the lecture "Clinical Biostatistics" is to give students an introduction to the most important statistical methods used in different areas of clinical research. First, quantifying clinical measurement error is central to the analysis of diagnostic studies and the assessment of agreement. Second, the randomized controlled trial (RCT) is the key concept to assess therapy, and advanced statistical methods are used in the design, implementation and analysis of RCTs. Finally, techniques for meta- analysis will be discussed. The following topics will be addressed: Confidence intervals for proportions, analysis of diagnostic studies, ROC curves, analysis of agreement, randomized controlled trials, hypothesis tests and sample size calculation, randomization and blinding, analysis of continuous and binary outcomes, multiplicity, subgroup analysis, protocol deviations, some special designs (crossover, equivalence, and cluster-randomized trials), principles of survival analysis, meta-analysis.

- Matthews, J. N. S. (2006). Introduction to Randomized Controlled Clinical Trials. Chapman & Hall/CRC Texts in Statistical Science.

- Pepe, M. (2003). The Statistical Evaluation of Medical Tests for Classification and Prediction. Oxford University Press

The module "Good Statistical Practice" introduces students to the specific concepts of research integrity that are related to the practice of statistics. This includes ethical principles of statistical practice, reproducibility, good written, visual and oral communication, good computational practice and computational efficiency. Besides theoretical insights into the foundations of statistical practice students acquire practical skills using modern computational tools (e.g. dynamic reporting, version control, containerization), they practice effective presentations and report writing of statistical results and they start to learn about general programming techniques (e.g. unit tests, debugging, job scripting, parallelization). The module provides insights and practical tools that are useful throughout the curriculum of the master program in biostatistics or other programs focused on quantitative research. It is taught in a flipped classroom setting: students are required to learn about concepts using provided material and complete assignments before an in-person session. Assignments and the in-person session contain peer and staff feedback and assessment.

We focus on the statistical analysis of health data which are collected in observational settings such as case-control or cohort studies. The most relevant measures of effect (risk, odds and rate ratios) are introduced, and methods for adjusting for confounders (Mantel-Haenszel and regression approaches) are thoroughly discussed. Specialized methods such as propensity score adjustments and conditional logistic regression are introduced for the analysis of matched data. Advanced topics such as causal inference with graphical models, imputation and measurement error and are also covered.

- Jewell, N. P. (2004): Statistics for Epidemiology. Chapman & Hall/CRC.

- Faraway, J. (2016): Extending the Linear Model with R. Chapman & Hall/CRC.

The analysis of survival times, or in more general terms, the analysis of time to event variables is concerned with models for censored observations. Because we cannot always wait until the event of interest actually happens, the methods discussed here are required for an appropriate handling of incomplete observations where we only know that the event of interest did not happen within a certain period. Most prominently, survival analysis is used in randomised clinical trials in oncology, but also in epidemiology, and also outside biostatistics in economics (duration analysis) or sociology (event history analysis). During the course, we will study the most important methods and models for censored data, including

- general concepts of censoring,
- simple summary statistics,
- estimation of survival curves,
- frequentist inference for two and more groups, and
- regression models for censored observations.

In the Biostatistics Journal Club biostatistical aspects of recent research papers or monographs are presented by each of the students and discussed together. All students together write a manuscript containing short summaries of all presentations.

For the Statistical Consulting module students will work under supervision on selected projects from the statistical consulting service of the Division of Biostatistics, they will write a reproducible report and present the results orally (see examples here). If no appropriate project is available or if there are other reasons the consulting project may be substituted by a term paper.

The master’s thesis is an independent research activity, which can, for example, be in the framework of an integrative project involving participants from other disciplines. It involves approximately a full-time 6 month workload and is concluded by a written report. A professor, who defines the subject and specifies the submission date, supervises the thesis. It is advisable to start choosing a topic for the master’s thesis in the second semester and to complete all compulsory coursework and as much elective coursework as possible before the thesis. See abstracts of examples here and a list of thesis titles here.

The master’s exam consists of an oral presentation of the master’s thesis followed by questions from an expert audience including the supervisor. The student needs to show the ability to clearly present the relevance of the thesis and to defend it in view of critical questions.